北京生物醫(yī)學(xué)工程

基于腦電信號(hào)的情緒識(shí)別

Emotion recognition based on electroencephalographsignals

作者：李賢哲暴偉謝能剛

單位：安徽工業(yè)大學(xué)管理科學(xué)與工程學(xué)院（安徽馬鞍山 243002） 通信作者：謝能剛，教授。E-mail：[email protected]

關(guān)鍵詞：腦電；視頻實(shí)驗(yàn)設(shè)計(jì)；特征提取；情緒識(shí)別

分類(lèi)號(hào)：R318.04

出版年·卷·期（頁(yè)碼）：2022·41·1（8-16）

摘要：

目的對(duì)多種情緒進(jìn)行快速準(zhǔn)確的識(shí)別，是目前腦機(jī)接口和情感計(jì)算領(lǐng)域的研究熱點(diǎn)。本文針對(duì)多情緒分類(lèi)問(wèn)題及個(gè)體差異的影響因素，設(shè)計(jì)一種電影片段誘發(fā)實(shí)驗(yàn)，利用機(jī)器學(xué)習(xí)算法對(duì)9位被試不同情緒的腦電信號(hào)進(jìn)行分析，以期能夠快速準(zhǔn)確識(shí)別不同被試的情緒狀態(tài)。方法首先采用非侵入式腦電設(shè)備收集被試在恐懼、憤怒、悲傷和快樂(lè)4種情緒下的腦電信號(hào)，通過(guò)對(duì)信號(hào)進(jìn)行降噪處理，使用時(shí)域、頻域和非線(xiàn)性動(dòng)力學(xué)的特征提取方法，共提取出15種不同的有效特征，并根據(jù)均方根特征和三維時(shí)域特征的散點(diǎn)圖來(lái)驗(yàn)證4種情緒之間的區(qū)分性；最后以平均準(zhǔn)確率作為分類(lèi)識(shí)別的評(píng)價(jià)指標(biāo)，應(yīng)用K近鄰算法對(duì)9位被試整體的腦電特征進(jìn)行訓(xùn)練和分類(lèi)。結(jié)果 3種時(shí)域特征的識(shí)別率差異較大，一階差分絕對(duì)值的均值特征平均準(zhǔn)確率達(dá)到95%，其余時(shí)域特征分類(lèi)效果一般；頻域特征中使用Welch法得到Gamma頻帶特征識(shí)別效果最好，平均準(zhǔn)確率超過(guò)95%；非線(xiàn)性動(dòng)力學(xué)特征識(shí)別率較好，平均分類(lèi)準(zhǔn)確率都超過(guò)90%。結(jié)論利用Welch法得到的Gamma頻帶特征和一階差分絕對(duì)值的均值作為最優(yōu)特征，能夠快速準(zhǔn)確識(shí)別不同被試的情緒狀態(tài)。

Objective Rapid and accurate recognition of multiple emotions is becoming a research hot spot in the field of brain-computer interface and affective computing currently. This paper focuses on a variety of emotional classification problems and the factors that cause individual differences, We design a kind of film segment inducing experiment. It can use machine learning algorithms to analysis the electroencephalograph（EEG） signals of 9 participants with different emotions and identify the emotional state of different participants quickly and accurately. Methods Firstly, it uses the non-invasive EEG equipment to collect the participants' EEG signals under the four emotions of fear, anger, sadness and happiness. Then, it uses time domain, frequency domain and nonlinear dynamic feature extraction methods to extract a total of 15 different effective features through the signal noise reduction processing. According to the root mean square feature and the scatter plot of the three-dimensional time domain feature, it can verify the distinction between the four emotions. Finally, it takes the average accuracy as the evaluation index of classification and recognition. Meanwhile, the k-nearest neighbor algorithm is applied to train and classify the overall EEG features of 9 participants. Results The recognition rates of the three time-domain features are quite different, the average accuracy of mean value of absolute value of first order difference features reaches 95%, but the classification effect of other time-domain features is general. In the frequency domain features, the Gamma frequency band feature recognition effect obtained by using the Welch method is the best, with an average accuracy rate exceeding 95%. The recognition rate of nonlinear dynamic features is good, and the average classification accuracy rate exceeds 90%. Conclusions Using the Gamma frequency band feature obtained by Welch method and the average accuracy of mean value of absolute value as the optimal feature, it can identify the emotional state of different participants quickly and accurately.

參考文獻(xiàn)：

[1]?Picard RW. Affective Computing[M]. Cambridge：MIT Press, 1997.

[2]?Othman M, Wahab A, Karim I, et al., EEG emotion recognition based on the dimensional models of emotions[J]. Procedia-Social and Behavioral Sciences, 2013,?97(2) : 30-37.

[3]?Ekman P. An argument for basic emotions[J]. Cognition & emotion,1992 , 6(3-4) : 169-200.

[4]?鄒吉林,張小聰,張環(huán),等. 超越效價(jià)和喚醒——情緒的動(dòng)機(jī)維度模型述評(píng)[J]. 心理科學(xué)進(jìn)展, 2011, 19(09) : 1339-1346.

?Zou JL, Zhang XC, Zhang H,et al. Beyond dichotomy of valence and arousal:review of the motivational dimensional model of affect[J]. Advances in Psychological Science, 2011, 19(9) : 1339-1346.

[5]?Wyczesany M, Ligeza TS. Towards a constructionist approach to emotions:verification of the three-dimensional model of affect with EEG-independent component analysis[J]. Experimental brain research, 2015, 233(3) : 723-733.

[6]?Tarnowski P, Ko?odziej M, Majkowski A, et al. Emotion recognition using facial expressions[J]. Procedia Computer Science,2017,108:1175-1184.

[7]?李曉慶,李東,蔣重清.?淺談?wù)Z音情緒識(shí)別的腦機(jī)制研究進(jìn)展[J].?吉林省教育學(xué)院學(xué)報(bào),?2018,?34(4)?:?166-168.

?Li XQ, Li D,Jiang CQ. A brief talk on the research progress of the brain mechanism of speech emotion recognition[J].?Journal of Educational Institute of Jilin Province,2018,?34(4)?:?166-168.

[8]?Sapiński T, Kamińska D, Pelikant A, et al. Emotion recognition from skeletal movements[J]. Entropy, 2019, 21(7) : 646.

[9]?Hsu YL, Wang J S, Chiang W C, et al. Automatic ecg-based emotion recognition in music listening[J].IEEE Transactions on Affective Computing,?2017, 11(1): 85-99.

[10]?Witvliet CVO, Vrana SR.Play it again Sam : Repeated exposure to emotionally evocative music polarises liking and smiling responses, and influences other affective reports,facial EMG, and heart rate[J]. Cognition and Emotion, 2007, 21(1) : 3-25.

[11]?Lee CK, Yoo SK, Park YJ, et al. Using neural network to recognize human emotions from heart rate variability and skin resistance[C]//2005 IEEE Engineering in Medicine and Biology 27th Annual Conference.Shanghai： IEEE,2006:5523-5525.

[12]?張冠華,余旻婧,陳果,等.?面向情緒識(shí)別的腦電特征研究綜述[J].中國(guó)科學(xué)：信息科學(xué),?2019,49（9）：1097-1118.

?Zhang GH,Yu WJ,Chen G,et al. Review of EEG features for emotion recognition[J]. Science China：?Information Sciences, 2019,49（9）：1097-1118.

[13]?陳景霞,王麗艷,賈小云,等.?基于深度卷積神經(jīng)網(wǎng)絡(luò)的腦電信號(hào)情感識(shí)別[J].?計(jì)算機(jī)工程與應(yīng)用, 2019,55(18):103-110.?

?Chen J X,Wang L Y,Jia X Y,et al. Emotion recognition of EEG signal based on deep convolutional neural network[J]. Computer Engineering and Applications, 2019,55(18):103-110.

[14]?Shi M, Wang C, Li X Z, et al. EEG signal classification based on SVM with improved squirrel search algorithm[J/OL]. Biomedical Engineering/Biomedizinische Technik, 2020.?[2020-09-29].https://doi.org/10.1515/bmt-2020-0038.

[15]?Li M, Xu H, Liu X, et al. Emotion recognition from multichannel EEG signals using K-nearest neighbor classification[J]. Technology and health care, 2018, 26(S1): 509-519.

[16]?聶聃,王曉韡,段若男,等.基于腦電的情緒識(shí)別研究綜述[J].?中國(guó)生物醫(yī)學(xué)工程學(xué)報(bào),?2012,31(4)?:?595-606.

?Nie D,Wang XW,Duan RN,et al.Review of Emotion Recognition Research Based on EEG[J]. Chinese Journal of Biomedical Engineering, 2012,31(4) : 595-606.

[17]?Kale RU,Ingale PM,Murade RT,et al.Comparison of quality power spectrum estimation (Bartlett, Welch, Blackman & Tukey) methods[J].?International Journal of Science and Modern Engineering,2013,1(5):28-31.

[18]?Djuric PM, Kay SM,Vijay KM, et al.Spectrum estimation and modeling[M]//?Madisetti?VK,Williams?DB.?Boca Raton: CRC Press LLC, 1999.

[19]?王健,吳勤衛(wèi),張新慧,等.?基于IDT90E36A的諧波表設(shè)計(jì)方法研究[J].?電氣傳動(dòng)自動(dòng)化,?2016,?38(6):37-40.

?Wang J,Wu QW,Zhang XH,et al. Research?on the design method of harmonic meter based on IDT90E36A[J]. Electric Drive Automation, 2016, 38(6):37-40.

[20]?劉志鵬.?認(rèn)知無(wú)線(xiàn)電中的寬頻段頻譜感知器設(shè)計(jì)[J].?信息通信,?2015(6):47-50.

?Liu ZP. Design of broadband spectrum perceptron in cognitive radio[J]. Information and Communications, 2015(6):47-50.

[21]?王晴.基于腦電信號(hào)的情緒分類(lèi)[D].北京：北京郵電大學(xué),2019.

?Wang Q. Emotion classification based on EEG signals[D]. Beijing：Beijing University of Posts and Telecommunications,2019.

[22]?楊曉明,晉玉劍,李永紅.經(jīng)典功率譜估計(jì)Welch法的MATLAB仿真分析[J].電子測(cè)試,?2011(7):101-104.

?Yang XM,Jin YJ,Li YH. MATLAB simulation analysis of classical power spectrum estimation Welch method[J]. Electronic Test, 2011(7):101-104.

[23]?張艷艷.近似熵理論及應(yīng)用[J].中國(guó)醫(yī)學(xué)物理學(xué)雜志,2009,26(6):1543-1546.

?Zhang YY.Theory and application of approximate entropy[J].Chinese Journal of Medical Physical,2009,26(6):1543-1546.

[24]?趙麗,萬(wàn)柏坤,綦宏志,等.近似熵應(yīng)用于老年性癡呆患者腦電研究[J].北京生物醫(yī)學(xué)工程,2003，22(2):140-142.

?Zhao L,Wang BK,Qi HZ,et al. Approximate entropy applied to the study of EEG in patients with Alzheimer's disease[J]. Beijing Biomedical Engineering, 2003，22(2):140-142.

[25]?劉慧,和衛(wèi)星,陳曉平.生物時(shí)間序列的近似熵和樣本熵方法比較[J].儀器儀表學(xué)報(bào), 2004,25(4): 806-807.

?Liu H,He WX,Chen XP.Comparison of approximate entropy and sample entropy of biological time series[J].Chinese Journal of Scientific Instrument, 2004,25(4): 806-807.

[26]?Kramer O. Dimensionality reduction with unsupervised nearest neighbors[M]. Berlin: Springer, 2013: 13-23.

[27]?黨杰.基于腦電的情緒識(shí)別與通道優(yōu)化選擇研究[D].蘭州：蘭州大學(xué),2019.

?Dang J.Research on emotion recognition and channel optimization selection based on EEG[D].Lanzhou：Lanzhou University,2019.

[28]?湯明宏. 基于腦電信號(hào)的情感識(shí)別研究[D].南京：南京郵電大學(xué),2018.

?Tang MH.Research?on emotion recognition based on EEG?signal[D].Nanjing：Nanjing University of Posts and Telecommunications,2018.

服務(wù)與反饋：

【文章下載】【加入收藏】

提示：您還未登錄，請(qǐng)登錄！點(diǎn)此登錄

51黑料吃瓜在线观看,51黑料官网|51黑料捷克街头搭讪_51黑料入口最新视频